---
title: "Splitting Complex English Sentences — Lee 2017"
description: "Neither approach shows which constructs cause trouble. 17 references, cited by 3 papers here."
authors: ["John Lee", "J. Buddhika K. Pathirage Don"]
year: 2017
venue: "Proceedings of the 15th International Conference on Parsing Technologies"
source: "https://aclanthology.org/W17-6307/"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/lee-2017-splitting-complex-english-sentences/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/lee-2017-splitting-complex-english-sentences/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2017](https://md.research.elamur.ai/free/en/papers/2017/page/4/) / Splitting Complex English Sentences

# Splitting Complex English Sentences

**Authors:** John Lee and J. Buddhika K. Pathirage Don · **Year:** 2017 · **Venue:** [Proceedings of the 15th International Conference on Parsing Technologies](https://md.research.elamur.ai/free/en/venues/international-workshop-conference-on-parsing-technologies/) ([International Workshop/Conference on Parsing Technologies 2017](https://md.research.elamur.ai/free/en/venues/international-workshop-conference-on-parsing-technologies/2017/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/W17-6307/) · [PDF](https://aclanthology.org/W17-6307.pdf) · [OpenAlex](https://openalex.org/W2760778844)

**Citations:** 5 citations (OpenAlex, 2026-10-14) · 3 in our library, of the 141,840 papers whose bibliographies we read

## Summary
The paper looks at one step of syntactic simplification: deciding which text span in a long English sentence should be cut out so it can become a separate sentence. The authors note that most earlier systems were judged by human ratings or by overall metrics such as BLEU. Neither approach shows which constructs cause trouble. To fix this, they built a public dataset from Wikipedia–Simple Wikipedia alignments in which one complex sentence corresponds to two simpler ones. A manually annotated test set marks the removed span, its construct type, whether the order changed, and the referring expressions. The training portion was cleaned automatically.

They compare two systems. The first is a baseline that matches hand-written dependency patterns over Stanford parser output. The second is a scikit-learn decision tree that filters those candidates using POS, dependency and span-length features. The decision tree raises F-measure from 0.49 to 0.67, mostly by improving precision. Results are broken down by construct, which gives a more detailed picture of the task than earlier evaluations.

## Key points
- Source data: 167,689 aligned normal/simple Wikipedia sentence pairs (Kauchak, 2013), of which 23,715 are one-to-two alignments. No alignment splits one sentence into more than two.
- Test set: 1,070 triplets (S, S1, S2), annotated in XML with the removed span, construct type, a `dest` attribute for reordering, and referent tags. Training triplets are kept only if at least 75% of the words of S1 or S2 occur in S. Their construct type is assigned by the baseline.
- Baseline: Stanford parser output, searched breadth-first for patterns such as conj/ccomp, acl:relcl, acl, appos, advcl, xcomp and parataxis. It removes the first match, at most one span per sentence. Scores: P 0.34, R 0.88, F 0.49.
- Decision tree scores: P 0.63, R 0.72, F 0.67. Its features cover the word's tag, the parent's tag, whether the parent is root, sibling and child tags, a comma between word and head, a DT-tagged subject, and span length.
- By construct, punctuation/parataxis is easiest (P 0.92, R 0.95) and appositive phrases are hardest (P 0.36, R 0.56). Many errors trace back to parser mistakes in clause attachment and clause boundaries.

## Interesting ideas
- Generate candidates with deliberately broad, high-recall tree patterns, then let a small learned classifier remove false positives; precision gains come without hand-tuning every rule.
- The useful learned rules are linguistically readable: conj counts as clausal only when the head is root, no split happens without a comma before an xcomp, and infinitives (TO) block splitting.
- Turn noisy parallel corpora into training data by keeping only one-to-two alignments with high word overlap, and let an unsupervised baseline supply the construct labels.
- Report results per construct instead of as one holistic score, and treat precision as a lower bound when gold labels reflect editors' choices rather than strict correctness.
- The annotation tracks the referent in S and its re-expression in each new sentence (for example "the man" → "He"), along with sentence order, which supports later evaluation of regeneration.

## For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

- **I1** (frames): candidate embedded clauses (participial, relative, appositive) could be proposed by loose UD patterns and then filtered, so each extracted sub-event carries both the pattern that fired and the filter's reason for keeping it.
- **I2** (tree): such checks can serve as explainable sanity tests that separate NP coordination from clause coordination before frames are built on the UD tree.
- **I3** (learn): the same overlap-filter-plus-weak-labeler recipe could produce silver data for clause segmentation, with a manually checked test slice kept separate from it.
- **I4** (learn): evaluation reports could break down by construction type and flag "acceptable but not in gold" outputs, so the system is not penalised for valid alternatives.
- **I5** (back): when the compressor or generator rewrites one complex sentence as several simple ones, these triplets could test whether referring expressions and event order are restored correctly.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

Also on the byline: [Applied Science and Technology Research Institute](https://ror.org/03xmkea05).

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 9,085 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 8,888 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 9,296 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 9,206 works

🧭 Topics (1): [English](https://md.research.elamur.ai/free/en/topics/english/) · 1,373 works

## Cite

Cite as: John Lee and J. Buddhika K. Pathirage Don. 2017. Splitting Complex English Sentences. Proceedings of the 15th International Conference on Parsing Technologies.

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{lee2017splitting,
  title = {{Splitting Complex English Sentences}},
  author = {John Lee and J. Buddhika K. Pathirage Don},
  booktitle = {Proceedings of the 15th International Conference on Parsing Technologies},
  year = {2017},
  url = {https://aclanthology.org/W17-6307/}
}
```

- [← previous A discourse account of intervention phenomena: An investigation of interrogatives](https://md.research.elamur.ai/free/en/papers/riches-2017-discourse-account-intervention-phenomena-investigation/)
- [next → The impact of flipped classroom model on teaching English grammar at 10th grade of high school](https://md.research.elamur.ai/free/en/papers/kheirabadi-2017-impact-flipped-classroom-model-teaching/)
- [↑ back to papers of 2017](https://md.research.elamur.ai/free/en/papers/2017/page/4/)

## References

17 references, 6 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Sandra Aluisio, Lucia Specia, Caroline Gasperin, and Carolina Scarton. 2010. [Readability Assessment for Text Simplification](https://md.research.elamur.ai/free/en/papers/aluisio-2010-readability-assessment-text-simplification/). In Proc. 5th Workshop on Innovative Use of NLP for Building Educational Applications. pages 1–9.
2. Advaith Siddharthan. 2002a. An Architecture for a Text Simplification System. In Proc. Language Engineering Conference (LEC). [find](https://scholar.google.com/scholar?q=Advaith+Siddharthan%2E+2002a%2E+An+Architecture+for+a+Text+Simplification+System%2E+In+Proc%2E+Language+Engineering+Conference+%28LEC%29%2E)
3. Advaith Siddharthan. 2002b. Resolving attachment and clause boundary ambiguities for simplifying relative clause constructs. In Proc.Student Workshop, ACL. [find](https://scholar.google.com/scholar?q=Advaith+Siddharthan%2E+2002b%2E+Resolving+attachment+and+clause+boundary+ambiguities+for+simplifying+relative+clause+constructs%2E+In+Proc%2EStudent+Workshop%2C+ACL%2E)
4. Sandra Aluı́sio, Lucia Specia, T. A. Pardo, E. G. Maziero, and R. P. Fortes. 2008. Towards Brazilian Portuguese Automatic Text Simplification Systems. In Proc. 8th ACM Symposium on Document Engineering. [find](https://scholar.google.com/scholar?q=Sandra+Alu%C4%B1%CC%81sio%2C+Lucia+Specia%2C+T%2E+A%2E+Pardo%2C+E%2E+G%2E+Maziero%2C+and+R%2E+P%2E+Fortes%2E+2008%2E+Towards+Brazilian+Portuguese+Automatic+Text+Simplification+Systems%2E+In+Proc%2E+8th+ACM+Symposium+on+Document+Engineerin)
5. M. A. Angrosh, Tadashi Nomoto, and Advaith Siddharthan. 2014. Lexico-syntactic text simplification and compression with typed dependencies. In Proc. COLING. [find](https://scholar.google.com/scholar?q=M%2E+A%2E+Angrosh%2C+Tadashi+Nomoto%2C+and+Advaith+Siddharthan%2E+2014%2E+Lexico%2Dsyntactic+text+simplification+and+compression+with+typed+dependencies%2E+In+Proc%2E+COLING%2E)
6. Advaith Siddharthan. 2011. Text Simplification using Typed Dependencies: A Comparison of the Robustness of Different Generation Strategies. In Proc. 13th European Workshop on Natural Language Generation. [find](https://scholar.google.com/scholar?q=Advaith+Siddharthan%2E+2011%2E+Text+Simplification+using+Typed+Dependencies%3A+A+Comparison+of+the+Robustness+of+Different+Generation+Strategies%2E+In+Proc%2E+13th+European+Workshop+on+Natural+Language+Generati)
7. J. De Belder and M. F. Moens. 2010. Text Simplification for Children. In Proc. SIGIR Workshop on Accessible Search Systems. [find](https://scholar.google.com/scholar?q=J%2E+De+Belder+and+M%2E+F%2E+Moens%2E+2010%2E+Text+Simplification+for+Children%2E+In+Proc%2E+SIGIR+Workshop+on+Accessible+Search+Systems%2E)
8. Advaith Siddharthan. 2014. A Survey of Research on Text Simplification. International Journal of Applied Linguistics 165(2):259–298. [↗](https://doi.org/10.1075/itl.165.2.06sid)
9. Stefan Bott, Horacio Saggion, and David Figueroa. 2012. A Hybrid System for Spanish Text Simplification. In Proc.Workshop on Speech and Language Processing for Assistive Technologies. [find](https://scholar.google.com/scholar?q=Stefan+Bott%2C+Horacio+Saggion%2C+and+David+Figueroa%2E+2012%2E+A+Hybrid+System+for+Spanish+Text+Simplification%2E+In+Proc%2EWorkshop+on+Speech+and+Language+Processing+for+Assistive+Technologies%2E)
10. Advaith Siddharthan and M. A. Angrosh. 2014. [Hybrid Text Simplification using Synchronous Dependency Grammars with Hand-written and Automatically Harvested Rules](https://md.research.elamur.ai/free/en/papers/siddharthan-2014-hybrid-text-simplification-synchronous-dependency/). In Proc. EACL.
11. R. Chandrasekar, Christine Doran, and B. Srinivas. 1996. [Motivations and Methods for Text Simplification](https://md.research.elamur.ai/free/en/papers/chandrasekar-1996-motivations-methods-text-simplification/). In Proc. COLING.
12. Gabriel Stanovsky and Ido Dagan. 2016. Annotating and Predicting Non-Restrictive Noun Phrase Modifications. In Proc. ACL. [find](https://scholar.google.com/scholar?q=Gabriel+Stanovsky+and+Ido+Dagan%2E+2016%2E+Annotating+and+Predicting+Non%2DRestrictive+Noun+Phrase+Modifications%2E+In+Proc%2E+ACL%2E)
13. David Vickrey and Daphne Koller. 2008. Sentence Simplification for Semantic Role Labeling. In Proc. ACL. [find](https://scholar.google.com/scholar?q=David+Vickrey+and+Daphne+Koller%2E+2008%2E+Sentence+Simplification+for+Semantic+Role+Labeling%2E+In+Proc%2E+ACL%2E)
14. Sanja Štajner, Maja Popović, Horacio Saggion, Lucia Specia, and Mark Fishel. 2016. Shared task on quality assessment for text simplification. In Proc. LREC. [find](https://scholar.google.com/scholar?q=Sanja+S%CC%8Ctajner%2C+Maja+Popovic%CC%81%2C+Horacio+Saggion%2C+Lucia+Specia%2C+and+Mark+Fishel%2E+2016%2E+Shared+task+on+quality+assessment+for+text+simplification%2E+In+Proc%2E+LREC%2E)
15. Sander Wubben, Antal van den Bosch, and Emiel Krahmer. 2012. [Sentence Simplification by Monolingual Machine Translation](https://md.research.elamur.ai/free/en/papers/wubben-2012-sentence-simplification-monolingual-machine-translation/). In Proc. ACL.
16. Wei Xu, Chris Callison-Burch, and Courtney Napoles. 2015. [Problems in current text simplification research: New data can help](https://md.research.elamur.ai/free/en/papers/xu-2015-problems-current-text-simplification-research/). Transactions of the Association for Computational Linguistics 3:283–297. [↗](https://doi.org/10.1162/tacl_a_00139)
17. Zhemin Zhu, Delphine Bernhard, and Iryna Gurevych. 2010. [A Monolingual Tree-based Translation Model for Sentence Simplification](https://md.research.elamur.ai/free/en/papers/zhu-2010-monolingual-tree-based-translation-model/). In Proc. ACL.

*Updated 24 October 2026*